https://aiweekly.co/alerts/anthropics-andy-hall-ai-auditors-hit-15-false-alarms-at-99 AI Weekly en 2026-11-08T14:01:59+00:00 Anthropic's Andy Hall: AI Auditors Hit 15% False Alarms at 99% https://aiweekly.co/alerts/semianalysis-visits-chahar-22-ai-campuses-3mmw-capex AI Weekly en 2026-11-08T12:34:06+00:00 SemiAnalysis Visits Chahar: 22 AI Campuses, <$3M/MW Capex https://aiweekly.co/alerts/recsa-stretches-qwen3-4b-to-2m-tokens-at-logarithmic-cost AI Weekly en 2026-11-08T12:12:02+00:00 RecSA stretches Qwen3-4B to 2M tokens at logarithmic cost https://aiweekly.co/alerts/nature-review-sorts-llms-into-four-social-science-roles AI Weekly en 2026-11-08T09:12:01+00:00 Nature review sorts LLMs into four social-science roles https://aiweekly.co/alerts/nature-review-casts-llms-in-four-roles-across-social-science AI Weekly en 2026-11-08T09:12:01+00:00 Nature review casts LLMs in four roles across social science https://aiweekly.co/alerts/nature-review-casts-llms-in-four-social-science-roles AI Weekly en 2026-11-08T09:12:01+00:00 Nature review casts LLMs in four social-science roles https://aiweekly.co/alerts/nature-human-behaviour-maps-llms-four-social-science-roles AI Weekly en 2026-11-08T09:12:01+00:00 Nature Human Behaviour maps LLMs' four social-science roles https://aiweekly.co/alerts/ml-beats-copenhagen-call-takers-at-spotting-stroke-calls AI Weekly en 2026-11-08T09:12:01+00:00 ML beats Copenhagen call-takers at spotting stroke calls https://aiweekly.co/alerts/copenhagen-ml-spots-stroke-on-helpline-calls-better-than-humans AI Weekly en 2026-11-08T09:12:01+00:00 Copenhagen ML spots stroke on helpline calls better than humans https://aiweekly.co/alerts/copenhagen-ml-model-spots-strokes-on-helpline-calls-at-63 AI Weekly en 2026-11-08T09:12:01+00:00 Copenhagen ML model spots strokes on helpline calls at 63% https://aiweekly.co/alerts/copenhagen-ml-caught-63-of-stroke-calls-beating-call-takers AI Weekly en 2026-11-08T09:12:01+00:00 Copenhagen ML caught 63% of stroke calls, beating call-takers https://aiweekly.co/alerts/copenhagen-ml-pipeline-outperforms-call-takers-on-stroke-calls AI Weekly en 2026-11-08T09:12:01+00:00 Copenhagen ML pipeline outperforms call-takers on stroke calls https://aiweekly.co/alerts/meta-strategy-chief-dave-wehner-joins-fireworks-as-cfo AI Weekly en 2026-11-08T08:00:05+00:00 Meta Strategy Chief Dave Wehner Joins Fireworks as CFO https://aiweekly.co/alerts/fts-tett-asks-why-america-hates-ai-as-asia-stays-bullish AI Weekly en 2026-11-08T05:08:49+00:00 FT's Tett asks why America hates AI as Asia stays bullish https://aiweekly.co/alerts/open-humanwam-open-sources-50k-hour-egocentric-robot-recipe AI Weekly en 2026-11-08T03:54:06+00:00 Open-HumanWAM open-sources 50K-hour egocentric robot recipe https://aiweekly.co/alerts/vissearch-bench-best-multimodal-agent-solves-2647-of-tasks AI Weekly en 2026-11-08T03:54:06+00:00 VisSearch-Bench: Best Multimodal Agent Solves 26.47% of Tasks https://aiweekly.co/alerts/openfrontiers-llms-solve-7-of-40-problems-without-the-web AI Weekly en 2026-11-08T03:54:06+00:00 OpenFrontiers: LLMs solve 7 of 40 problems without the web https://aiweekly.co/alerts/google-adk-2140-graduates-core-modules-adds-teams-runner AI Weekly en 2026-11-08T03:29:08+00:00 Google ADK 2.14.0 Graduates Core Modules, Adds Teams Runner https://aiweekly.co/alerts/apiiro-claude-code-dont-ask-again-leaks-keys-in-1122-repos AI Weekly en 2026-11-08T00:34:31+00:00 Apiiro: Claude Code 'Don't Ask Again' Leaks Keys in 1,122 Repos https://aiweekly.co/alerts/kv-cache-gate-holds-9014-ruler-at-75-compression-on-qwen3-8b AI Weekly en 2026-11-08T00:12:01+00:00 KV cache gate holds 90.14 RULER at 75% compression on Qwen3-8B https://aiweekly.co/alerts/openai-hires-hacktron-to-red-team-third-party-integrations AI Weekly en 2026-11-07T22:59:50+00:00 OpenAI Hires Hacktron to Red-Team Third-Party Integrations https://aiweekly.co/alerts/openais-ai-disproof-of-1998-eth-shocks-co-author-impagliazzo AI Weekly en 2026-11-07T18:39:20+00:00 OpenAI's AI Disproof of 1998 ETH Shocks Co-Author Impagliazzo https://aiweekly.co/alerts/elvis-estate-and-eminem-publisher-sue-suno-over-ai-voice-cloning AI Weekly en 2026-11-07T18:39:20+00:00 Elvis Estate and Eminem Publisher Sue Suno Over AI Voice Cloning https://aiweekly.co/alerts/dreamfree-skips-video-generation-for-11hz-robot-control AI Weekly en 2026-11-07T15:56:32+00:00 DreamFree Skips Video Generation for 11Hz Robot Control https://aiweekly.co/alerts/study-pretraining-seed-not-data-drives-jailbreak-holes AI Weekly en 2026-11-07T15:56:32+00:00 Study: pretraining seed, not data, drives jailbreak holes https://aiweekly.co/alerts/harness-s-trains-agent-policies-across-a-pool-of-sibling-harnesses AI Weekly en 2026-11-07T15:56:32+00:00 Harness-S trains agent policies across a pool of sibling harnesses https://aiweekly.co/alerts/nyt-us-cleared-nvidia-h200-firms-in-china-have-empty-offices AI Weekly en 2026-11-07T15:28:12+00:00 NYT: U.S.-Cleared Nvidia H200 Firms in China Have Empty Offices https://aiweekly.co/alerts/nous-researchs-arete-27b-cuts-qwen-propaganda-from-22-to-09 AI Weekly en 2026-11-07T15:28:12+00:00 Nous Research's Arete-27B Cuts Qwen Propaganda From 22% to 0.9% https://aiweekly.co/alerts/aws-sagemaker-splits-gpus-into-up-to-7-partitions-via-mig AI Weekly en 2026-11-07T14:01:00+00:00 AWS SageMaker Splits GPUs Into Up to 7 Partitions via MIG https://aiweekly.co/alerts/nasscom-chief-warns-ai-is-squeezing-indias-it-freshers-out AI Weekly en 2026-11-07T12:35:06+00:00 Nasscom chief warns AI is squeezing India's IT freshers out https://aiweekly.co/alerts/my-baby-bunny-ai-teddy-gave-unprompted-self-harm-reply-bbc-finds AI Weekly en 2026-11-07T12:12:02+00:00 My Baby Bunny AI teddy gave unprompted self-harm reply, BBC finds https://aiweekly.co/alerts/huawei-debuts-qingyun-hm650-as-first-100-domestic-pc AI Weekly en 2026-11-07T09:30:30+00:00 Huawei Debuts Qingyun HM650 as First '100% Domestic' PC https://aiweekly.co/alerts/nature-human-behaviour-maps-llms-four-roles-in-social-science AI Weekly en 2026-11-07T09:12:02+00:00 Nature Human Behaviour maps LLMs' four roles in social science https://aiweekly.co/alerts/nature-review-sorts-llms-into-four-social-science-roles AI Weekly en 2026-11-07T09:12:02+00:00 Nature review sorts LLMs into four social-science roles https://aiweekly.co/alerts/xu-evans-nature-review-four-llm-roles-in-social-science AI Weekly en 2026-11-07T09:12:02+00:00 Xu-Evans Nature review: four LLM roles in social science https://aiweekly.co/alerts/nature-human-behaviour-maps-llms-four-roles-in-social-science AI Weekly en 2026-11-07T09:12:02+00:00 Nature Human Behaviour maps LLMs' four roles in social science https://aiweekly.co/alerts/xu-evans-map-four-llm-roles-across-social-sciences AI Weekly en 2026-11-07T09:12:02+00:00 Xu, Evans map four LLM roles across social sciences https://aiweekly.co/alerts/copenhagen-study-ml-outperforms-call-takers-on-stroke-flags AI Weekly en 2026-11-07T09:12:02+00:00 Copenhagen study: ML outperforms call-takers on stroke flags https://aiweekly.co/alerts/ml-tops-copenhagen-helpline-call-takers-at-spotting-stroke AI Weekly en 2026-11-07T09:12:02+00:00 ML tops Copenhagen helpline call-takers at spotting stroke https://aiweekly.co/alerts/copenhagen-study-ml-beats-dispatchers-at-stroke-recognition AI Weekly en 2026-11-07T09:12:02+00:00 Copenhagen study: ML beats dispatchers at stroke recognition https://aiweekly.co/alerts/censys-finds-800-exposed-artex-ai-hacking-tools-online AI Weekly en 2026-11-07T07:57:32+00:00 Censys Finds 800+ Exposed Artex AI Hacking Tools Online https://aiweekly.co/alerts/ramp-enterprise-ai-spend-grew-15-as-anthropic-nears-ipo AI Weekly en 2026-11-07T07:57:32+00:00 Ramp: Enterprise AI Spend Grew 1.5% as Anthropic Nears IPO https://aiweekly.co/alerts/meta-adds-spaces-to-muse-agent-topic-tabs-with-app-hooks AI Weekly en 2026-11-07T07:57:32+00:00 Meta Adds Spaces to Muse Agent, Topic Tabs With App Hooks https://aiweekly.co/alerts/shenzhen-clears-smuggled-chip-stolen-iphone-stalls-before-apec AI Weekly en 2026-11-07T07:57:32+00:00 Shenzhen Clears Smuggled-Chip, Stolen-iPhone Stalls Before APEC https://aiweekly.co/alerts/anthropic-computer-science-dominates-research-use-of-claude AI Weekly en 2026-11-07T06:12:02+00:00 Anthropic: computer science dominates research use of Claude https://aiweekly.co/alerts/judgmentbench-pairwise-scoring-beats-rubrics-0908-to-0150 AI Weekly en 2026-11-07T06:12:02+00:00 JudgmentBench: pairwise scoring beats rubrics 0.908 to 0.150 https://aiweekly.co/alerts/dpdm-lifts-agent-accuracy-to-8492-across-50-benchmarks AI Weekly en 2026-11-07T03:53:45+00:00 DPDM lifts agent accuracy to 84.92% across 50 benchmarks https://aiweekly.co/alerts/speculative-listening-trims-full-duplex-voice-ai-lag-by-73 AI Weekly en 2026-11-07T03:53:45+00:00 Speculative Listening trims full-duplex voice AI lag by 73% https://aiweekly.co/alerts/tailsteer-tames-multi-turn-drift-across-flux-sensenova-qwen AI Weekly en 2026-11-07T03:53:45+00:00 TailSteer tames multi-turn drift across FLUX, SenseNova, Qwen https://aiweekly.co/alerts/helm-lifts-vla-robot-success-128pp-with-45x-less-memory AI Weekly en 2026-11-07T03:53:45+00:00 HeLM lifts VLA robot success 12.8pp with 4.5x less memory https://aiweekly.co/alerts/evomemstate-adds-rl-trained-reasoning-state-to-agent-memory AI Weekly en 2026-11-07T03:53:45+00:00 EvoMemState adds RL-trained reasoning state to agent memory https://aiweekly.co/alerts/openai-wires-11b-first-close-at-125t-below-14t-target AI Weekly en 2026-11-07T03:30:52+00:00 OpenAI Wires $11B First Close at $1.25T, Below $1.4T Target https://aiweekly.co/alerts/ricursives-ripple-65b-tops-bigger-us-open-weight-models AI Weekly en 2026-11-07T03:12:01+00:00 Ricursive's Ripple-65B tops bigger US open-weight models https://aiweekly.co/alerts/meta-ships-muse-wonder-trained-models-to-all-muse-meta-ai-users AI Weekly en 2026-11-07T02:02:08+00:00 Meta ships Muse Wonder-trained models to all Muse, Meta AI users https://aiweekly.co/alerts/google-takes-over-crusoes-wyoming-site-after-customer-miss AI Weekly en 2026-11-07T00:34:27+00:00 Google Takes Over Crusoe's Wyoming Site After Customer Miss https://aiweekly.co/alerts/copenhagen-ml-model-beats-call-takers-at-helpline-stroke-id AI Weekly en 2026-11-07T00:12:02+00:00 Copenhagen ML model beats call-takers at helpline stroke ID https://aiweekly.co/alerts/memoir-adds-1192pt-evidence-recall-to-llm-agent-memory AI Weekly en 2026-11-06T21:35:07+00:00 MEMOIR adds 11.92pt evidence recall to LLM agent memory https://aiweekly.co/alerts/vergecast-tech-billionaires-2026-midterm-bets-flopped AI Weekly en 2026-11-06T21:12:01+00:00 Vergecast: tech billionaires' 2026 midterm bets flopped https://aiweekly.co/alerts/hundreds-storm-microsofts-amsterdam-data-center-build-site AI Weekly en 2026-11-06T21:12:01+00:00 Hundreds storm Microsoft's Amsterdam data center build site https://aiweekly.co/alerts/ml-model-tops-copenhagen-call-takers-on-stroke-recognition AI Weekly en 2026-11-06T21:12:01+00:00 ML model tops Copenhagen call-takers on stroke recognition https://aiweekly.co/alerts/doj-indicts-three-in-129m-ai-ceo-voice-clone-fraud-ring AI Weekly en 2026-11-06T19:56:30+00:00 DOJ Indicts Three in $129M AI CEO Voice-Clone Fraud Ring https://aiweekly.co/alerts/microsoft-lets-2023-oracle-ai-cloud-deal-lapse-over-older-nvidia-chips AI Weekly en 2026-11-06T19:56:30+00:00 Microsoft Lets 2023 Oracle AI Cloud Deal Lapse Over Older Nvidia Chips https://aiweekly.co/alerts/anthropic-ipo-buyers-float-11t-valuation-after-ai-rout AI Weekly en 2026-11-06T19:56:30+00:00 Anthropic IPO Buyers Float $1.1T Valuation After AI Rout https://aiweekly.co/alerts/prix-goncourt-winner-philippe-jaenada-admits-chatgpt-use AI Weekly en 2026-11-06T18:12:02+00:00 Prix Goncourt winner Philippe Jaenada admits ChatGPT use https://aiweekly.co/alerts/newton-ai-backlash-has-a-mandate-but-lacks-a-congress-plan AI Weekly en 2026-11-06T18:12:02+00:00 Newton: AI backlash has a mandate but lacks a Congress plan https://aiweekly.co/alerts/hugging-face-publishes-70000-word-moe-training-handbook AI Weekly en 2026-11-06T18:12:02+00:00 Hugging Face publishes 70,000-word MoE training handbook https://aiweekly.co/alerts/tpot-automl-tool-rewritten-adds-multi-objective-search AI Weekly en 2026-11-06T18:12:02+00:00 TPOT AutoML tool rewritten, adds multi-objective search https://aiweekly.co/alerts/legoenv-builds-5465-agent-environments-from-reusable-atoms AI Weekly en 2026-11-06T18:12:02+00:00 LegoEnv builds 5,465 agent environments from reusable atoms https://aiweekly.co/alerts/ml-pipeline-beats-copenhagen-call-takers-on-stroke-recognition AI Weekly en 2026-11-06T18:12:02+00:00 ML pipeline beats Copenhagen call-takers on stroke recognition https://aiweekly.co/alerts/copenhagen-study-ml-lifts-stroke-call-recognition-to-63 AI Weekly en 2026-11-06T18:12:02+00:00 Copenhagen study: ML lifts stroke call recognition to 63% https://aiweekly.co/alerts/norris-paper-links-pleiades-myths-to-humans-leaving-africa AI Weekly en 2026-11-06T18:12:02+00:00 Norris paper links Pleiades myths to humans leaving Africa https://aiweekly.co/alerts/openssh-sshole-flaw-opens-musl-linux-accounts-to-attackers AI Weekly en 2026-11-06T17:06:31+00:00 OpenSSH '*sshole' flaw opens musl Linux accounts to attackers https://aiweekly.co/alerts/marmot-library-consortium-logs-2b-ai-bot-hits-in-30-days-cbc AI Weekly en 2026-11-06T17:06:31+00:00 Marmot library consortium logs 2B AI-bot hits in 30 days: CBC https://aiweekly.co/alerts/googles-muir-cuts-perplexity-36-vs-transformer-at-134b AI Weekly en 2026-11-06T17:06:31+00:00 Google's Muir cuts perplexity 3.6% vs Transformer at 1.34B https://aiweekly.co/alerts/tb-mathmodel-gpt-6-astra-tops-contest-bench-at-6458-o-eval AI Weekly en 2026-11-06T17:06:31+00:00 TB-MathModel: GPT-6 Astra tops contest bench at 64.58% O-Eval https://aiweekly.co/alerts/democratic-congress-lines-up-nvidia-chip-export-financing-probes AI Weekly en 2026-11-06T17:06:31+00:00 Democratic Congress Lines Up Nvidia Chip Export, Financing Probes https://aiweekly.co/alerts/tsmc-weighs-3d-dram-for-nvidia-as-ai-memory-shortage-bites AI Weekly en 2026-11-06T17:06:31+00:00 TSMC Weighs 3D DRAM for Nvidia as AI Memory Shortage Bites